How has unemployment in the UK changed over the past 100 years?

These are the words of a millwright – a skilled worker who installs and repairs machinery – who became unemployed in 1930, describing the dire conditions at the peak of the Great Depression. One in six British workers was unemployed during that period, with conservative estimates of the unemployment rate reaching 15.6% in 1932 (see Figure 1).

Since then, aggregate unemployment rates have improved somewhat. At its worst in the 1980s, only one in eight workers was unemployed, with the unemployment rate climbing to 11.9% in 1984. Since 1990, the unemployment rate has averaged less than 8%, and in 2011, after the global financial crisis of 2007-09, it reached only 8.5%.

Figure 1: The UK’s unemployment rate, 1920-2016

Source: Bank of England, unemployment rate in the UK [AURUKM], retrieved from FRED, Federal Reserve Bank of St. Louis.
Note: Data for interwar Britain is the adjusted unemployment rate from Feinstein, 1972.

To understand patterns in unemployment, we need to consider what we mean by the term, its various types and how we measure it.

First, we need to break unemployment into its constituent types. There is always some baseline level of unemployment in the economy, including frictional unemployment from workers moving between jobs or entering the workforce.

Beyond this baseline churn, some unemployment is cyclical, which means that it is associated directly with recessions. Unemployment rises during these periods through falling demand for products, increased uncertainty and cost-cutting measures.

This unemployment is called cyclical because it is driven by the business cycle, and it typically falls when the economy begins growing again. For individuals, of course, this unemployment is serious, and recent research shows how cyclical unemployment can have long-term ‘scarring’ effects on workers’ labour market prospects and other outcomes (Huckfeldt, 2022; Yagan, 2019; Rothstein, 2026).

The other major kind of unemployment is structural unemployment, which comes from differences between the kinds of jobs that are available and the kind of skills that workers have or the kind of jobs for which they are looking.

When workers traditionally conceive of having a career, they envision training in an occupation and continuing to work in that field, moving up through the ranks over the course of 40-plus years. But which industries are hiring and what kind of skills they are looking for can change significantly over the course of a decades-long career, with changes in technology, globalisation, productivity and consumer demand.

The changing composition of industries in the economy alters the demand for skills, and this can cause some workers’ skills to be misaligned with the jobs available. Structural unemployment can also have a geographical component, where there are places in the country or in the world where a worker’s skills are valued, just not in the community in which they currently live.

How can we understand and measure unemployment in history?

For most of British history, there was no such thing as an unemployment rate. People were out of work and sometimes desperately impoverished. But prior to the turn of the 20^th century, unemployment was not typically conceived of as a social problem that might require large-scale government intervention. Instead, unemployment was seen as a personal failure, indelibly linked with poverty and socially stigmatised (Hatton, 2004).

No aggregate unemployment rate was measured because there was not a sense in which unemployment could reflect a broader social or structural issue. Our ability to measure and track unemployment over time relies on having a stable conception of unemployment and consistent tools for monitoring it in the labour force, neither of which existed in Britain before the 20^th century.

Attitudes towards unemployment shifted at the turn of that century with the early social surveys (Booth, 1902) and the development of a better understanding of the cyclical and structural aspects of unemployment (Beveridge, 1909). Unemployment insurance was established in Britain in 1911 to address these cyclical aspects, and its expansion after the First World War to cover much of the workforce reflected a broader understanding of the structural aspects.

Claimant counts from the unemployment insurance programme were immediately used as aggregate measures of labour market performance in interwar Britain, making the UK an early leader in the collection of workforce statistics. The basis for these figures shifted again after the Second World War when the programme was made even more comprehensive.

Yet it was only in the 1980s that survey-based measures in line with modern definitions of unemployment emerged, following the development of the Labour Force Survey (LFS) and standard International Labour Organization (ILO) definitions of unemployment (Denman and McDonald, 1996).

While unemployment rates back to 1855 have been collected and made easily available, the varied definitions and methods used to establish these numbers, and especially the extremely narrow populations covered prior to 1911, hamstring any long-run comparison of these rates.

It is important to note that these definitional problems are not fully solved today. Our definition of the unemployment rate has again become too narrow to capture the labour market performance of the economy.

Unemployment rates do not include discouraged workers, who are no longer searching for work. Nor do they include some young workers who are ‘not in employment, education or training’ (NEETs) and may want to work but are not actively looking. Unemployment rates also miss underemployment – that is, those who are working fewer hours than they desire.

Further, this measure does not offer insight into those working in industries that do not align with their skills, or those who are in casual work or zero-hours contracts. All of these issues mean that the unemployment rate can understate weakness in the labour market.

These issues are not just modern. For example, one challenge when comparing men and women’s historic unemployment rates is that women were often only counted as in the labour force when they were employed, artificially lowering their reported unemployment rate (Paker, 2024).

What’s more, because the unemployment data for the UK were for so long linked to unemployment insurance, we have good disaggregated figures by industry and gender for the past century, but almost none by occupation, skill, age, education or ethnicity until the late 20^th century.

How has cyclical unemployment changed?

Imagine if, three years from now, the value of worldwide trade was cut in half and industrial production declined by a third. This was the situation from 1929 to 1932, when, coupled with a collapse of product prices, almost every major economy was thrown into the Great Depression and a period of high cyclical unemployment, including in the UK (Feinstein et al, 1997). No recession has matched its severity in the past century.

Central to the Great Depression was the gold standard – a monetary system in which nearly all currencies’ values were fixed to a specified quantity of gold. This arrangement was restored throughout the world in the 1920s in an effort to promote economic stability after the First World War.

The resumption of the gold standard forced deflationary policies in order to protect gold reserves. In the early 1930s, deflationary pressure was transmitted internationally, amplified by a lack of international cooperation and credibility that destabilised the system and contributed to the Great Depression (Eichengreen, 1996).

In the world of the gold standard, policy-makers could not actively fight a recession, as their hands were tied defending the exchange rate. Leaving the gold standard was key to the eventual economic recovery from the Great Depression for many countries (Eichengreen and Sachs, 1985; Lennard and Paker, 2026; Ellison et al, 2024).

Constraints and credibility still play a role in the management of more modern recessions. The 1970s sterling crisis that led to the International Monetary Fund (IMF) loan to the UK was associated with constraints, including cuts to government spending and deficit reductions. This dovetailed with the growing influence of monetarist ideas, ultimately leading to money supply targets. Under Margaret Thatcher’s government, which came to power in 1979, these targets were allowed to drive monetary policy, leading to the highest-ever level of the Bank of England’s policy interest rate, 17%, in 1979.

These high interest rates attracted foreign capital inflows, which pushed up the value of sterling, making UK exports less competitive. This squeezed some industries like manufacturing, while also reducing domestic demand, both of which led to unemployment. This all occurred in an environment of reduced government spending and increased taxation, which further reduced aggregate demand.

The head of Margaret Thatcher’s policy unit, John Hoskyns, wrote retrospectively that they had ‘accidentally engineered’ a major recession through money supply targets (Needham, 2015). Nevertheless, the credible commitment to reducing inflation may have been an important monetary policy regime shift that helped to anchor inflation expectations (Sargent, 2013).

In combating the global financial crisis of 2007-09, some lessons were learned. Policy-makers acted decisively to reduce constraints. They cut interest rates and injected liquidity into financial markets, stepped in to save banks and pursued fiscal stimulus programmes. Each of these interventions prevented a worse recession, even if more could have been done (Eichengreen, 2014).

In an obvious sense then, over the past century, our tools for responding to recessions – and thus moderating their cyclical unemployment costs – have improved. Monetary policy, when unconstrained by the gold standard or ideology, can help to counteract the impact and inject liquidity into financial markets. Fiscal policy has become a key tool for increasing demand and restoring consumer confidence. When used well, these seem to make recessions less severe.

But we should always be cautious about declaring these problems solved through good monetary policy or that we can find a way out of the cycle of ‘booms and busts’. This is short-sighted. The work of central banks and economists in governments is essential because new challenges constantly arise for which there are no established tools, rules or precedents. Recent examples include shadow banking, cryptocurrency and central bank digital currencies, and inflation arising from the pandemic.

How has structural unemployment changed?

Over the past century, the UK has experienced broad shifts in the locus of economic activity, having moved away from agriculture towards manufacturing, then shifting from manufacturing to services. Figure 2 shows these major shifts in the sectors in which workers are employed in the census from 1921 to 2021.

Figure 2: Change in employment share by sector, 1921-2021

Source: Census for England and Wales
Note: Figures for 1921 and 1931 include employed and unemployed workers over 12 and 14 years old, respectively, and are drawn from Table C of the 1931 Census Industry Report for England and Wales. Figures for 1961, 1971, 1991 and 2001 are drawn from the IPUMS international harmonised 1% microdata samples of the UK Census including only employed workers aged 16-74. Figures for 2011 draw from the Census aggregate tables covering employed workers 16-74 and figures for 2021 include employed workers over 16.

While the share of workers in manufacturing had already fallen substantially by the 1971 census from its 1921 level, the decline was especially dramatic through to 1991, decreasing by 8.8 percentage points or 22%. This decline from the 1970s to the 1990s is notable even in international comparisons (Muellbauer, 2022, see Figure 3). These patterns of deindustrialisation also continued through to 2011.

To the extent that deindustrialisation displaced manufacturing workers, it created structural unemployment, especially in periods in which the move towards a service-based economy occurred more rapidly. But some of these broad shifts happened at the margins, for example, as young people chose different careers and older workers retired.

Figure 3: Change in employment share by manufacturing industry, 1924-34

Source: Labour Gazette
Note: Change in employment share from July 1924 to July 1934. Only the ten manufacturing industries with the largest increases and the ten with the largest decreases in employment are shown, out of 69 industries.

These broad patterns of structural change also obscure important shifts in the composition of jobs within each major category. Deindustrialisation from manufacturing to services is not the only process that affects the industrial or occupational composition of work.

For example, Figure 2 might make it seem like there was not much change in manufacturing from 1921 to 1931. But this masks a major realignment from heavy to light manufacturing (as Figure 3 shows), which caused structural unemployment during the interwar period.

While some kinds of manufacturing, such as cotton textile manufacturing and steel, were in dramatic decline, other kinds, such as electrical, cars and artificial silk, were actually growing. The aggregate figures hide these structural shifts, which contributed to the high levels of unemployment between the First and Second World Wars (Paker, 2026).

These more disaggregated perspectives are necessary to understand more recent structural unemployment, which has largely happened within the services sector. Research on job polarisation highlights the rise of high- and low-skill occupations at the expense of middle-level routine occupations since the late 1980s, leading to the decline of occupations such as bank tellers and travel agents (Goos and Manning, 2007; Cristini et al, 2018; Paker, 2023).

While the effects of artificial intelligence (AI) on the labour market are still unpredictable, this new technology may cause even further structural shifts in the balance of industries and occupations within services (Eloundou et al, 2024; Brynjolfsson et al, 2025).

The composition of jobs in the economy is always changing – and has been for the past century. But in that time, we have developed a better understanding of three aspects of structural unemployment.

First, it is critical that workers can move between industries in response to changes. Many economists have studied the efficacy of various active labour market policies to facilitate these transitions (Le Barbanchon et al, 2024 provide an overview).

Second, and relatedly, we have a better understanding that the effects of structural change on labour markets are unequal throughout the workforce. The shifting types of jobs in the economy displace some workers and help others, and it is important to remember that these effects are often unpredictable.

For example, software programming was a focus of major initiatives for skill development, such as the Year of Code in 2014. But now, job openings have declined for software programming, especially for young people (National Foundation For Educational Research, NFER, 2025).

Finally, we are increasingly recognising that cyclical and structural unemployment interact and that worker reallocation across industries during economic downturns shapes aggregate unemployment (Şahin et al, 2014; Chodorow-Reich and Wieland, 2020).

Where can I find out more?

• The end of the gold standard and the beginning of the recovery from the Great Depression: VoxEU article by Jason Lennard and Meredith Paker.
• Lousy and lovely jobs: The rising polarization of work in Britain: Article in the Review of Economics and Statistics by Maarten Goos and Alan Manning.
• Job search, unemployment insurance, and active labor market policies: Chapter in the Handbook of Labor Economics.
• The Thatcher legacy: Lessons for the future of the UK economy: Chapter in the Economy 2023 inquiry.

Who are experts on this question?

• Meredith Paker
• Jason Lennard
• Tim Hatton
• Alan Manning

Author: Meredith Paker

This article is part of a series exploring economic and social changes since 1926, to mark the centenary of the Economic History Society.

Photo: Mike Beverley for iStock